By 'lag', I mean one port from each mc-lag peer connects to a port each on the uplink switch (consider this a single switch spine). The mc-lag is an active-active forwarding pair for L2 traffic (aruba's vsx solution). The reason I ask is that I have been told "If uplink is a lag, then it is not a route-only-port (i.e. not L3). The ebgp packet from remote neighbor can end up in any of the mc-lag peers. Hence ebgp multihop is required to be configured to 2, since the packet can go from North to sec, sec to pri. Similarly, it can go from north to pri to sec."
I don't understand how this works.
If its not a rop towards uplink, if dst is pri, if packet reaches sec, sec will not decrement ttl right ?
How does this lag work ? Are there two different IPs ? (i think not) Is it one logical SVI IP or virtual IP ? Do both respond to the ARP or does only the primary always respond ?
- They say because this is active/active, both the peers have independent BGP sessions to remote BGP peer.